The classification presented here is based on a multivariate analysis of the shield bosses from the Upper Thames region, carried out in 1975 (Dickinson 1976, 274–90, figs. 25–9). Intuitive methods had failed to work, just as they had probably failed for previous scholars, because shield bosses seem to lack strikingly obvious diagnostic features. By contrast, the adoption of an explicit and more broadly-based approach not only took advantage of the substantial sample, but also was well suited to the circumstances of the enquiry. Residual doubts, however, meant that the results were not published at the time, though others have found the system both useful and valid (e.g. Welch 1983, 136–40, where the essentials are summarized; Hirst 1985, 91) and Heinrich Härke (HH) also adopted it as a base for the dating of shield burials in the course of his research. There is little excuse, therefore, for withholding publication further, and it is presented here substantially as it was written then, save for some modifications and additions.

The classification was devised as an aid to dating male graves in the context of a cultural-historical study of the Upper Thames cemeteries. Its aim was chronological order, and its principal assumption was that similarity of form reflected a similarity in time and space of production, use and final deposition. It was largely innocent of developments in the theoretical ramifications of artefact classification (e.g. Hill and Evans 1972; but see now especially Miller 1985), and no doubt both contains and obscures categories which have little or nothing to do with time and space: some of these are explored by HH later in this volume.